Your browser is no longer supported

For the best possible experience using our website we recommend you upgrade to a newer version or another browser.

Your browser appears to have cookies disabled. For the best experience of this website, please enable cookies in your browser

We'll assume we have your consent to use cookies, for example so you won't need to log in each time you visit our site.
Learn more

Study linking HRT to breast cancer 'was wrong'

  • Comment

Several newspapers have reported that hormone replacement therapy (HRT) does not increase the risk of breast cancer.

The reports said that the study that reported the link in 2002 was “fundamentally flawed”.

The original research found that women who had taken HRT were twice as likely to develop breast cancer and more likely to die. The Daily Telegraph and other newspapers reported that these 2002 findings “triggered a collapse in confidence in hormone replacement therapy and the number of women taking it halved”.

Today’s news reports are based on a new analysis of the data from the Million Women Study and two other studies. The researchers wanted to determine whether the association between HRT and breast cancer actually showed that HRT caused an increased risk in breast cancer. They appraised each study against nine criteria that would need to be met before it was possible to say whether HRT caused breast cancer. For example, one criterion assessed was whether the women had taken HRT before they had breast cancer.

This new analysis did not suggest that the original studies were poorly performed but it did highlight limitations to what the studies can tell us.

The analysis found that all three studies failed to meet the majority of the criteria and therefore none of the studies was able to establish whether HRT increases the risk of breast cancer.

Where did the story come from?

The study was carried out by researchers from the University of Cape Town and UK universities including the University of Surrey and Imperial College London. It was published in the peer-reviewed Journal of Family Planning and Reproductive Health Care.

The newspapers reported that this analysis showed that the previous studies had not demonstrated causality.

What kind of research was this?

This was a reanalysis of three studies suggesting that HRT caused an increased risk of breast cancer, to see whether the association was actually causal. The three studies were:

  • The Collaborative Reanalysis, published in 1997, which had pooled data of 51 studies that looked at the risk of breast cancer in relation to HRT use.
  • The Women’s Health Initiative, which included two randomised controlled trials published in 2002. In the first, approximately 5,000 women received oestrogen-only HRT and 5,000 received a placebo for an average of 6.8 years. In the second study, more than 8,000 women received oestrogen plus progesterone HRT and more than 8,000 women received a placebo and were followed for an average of 5.2 years. Oestrogen-only HRT can only be given to women who have had a hysterectomy (they have no uterus). In women who have an intact uterus, giving “unopposed” oestrogen without progesterone can cause overgrowth of the lining of the uterus (the endometrium), which increases the risk of endometrial cancer.
  • Data from the Million Women Study, published in 2003. This was a prospective cohort study that followed more than a million UK women who were over 50. Women were invited to join the study when they received their invitation to attend breast cancer screening between 1996 and 2001.

For each of the three studies the researchers looked at whether they met several criteria needed to demonstrate causality, and presented their findings in a series of papers. Besides the causality criteria, the three study designs have the following inherent strengths and limitations:

  • Pooled analysis, such as the Collaborative Reanalysis study, is good to look for trends and consistency when studies have been performed by different research groups and in different cohorts. However, pooled analysis can be unreliable if the included studies are of poor quality or have inconsistent methodology.
  • Randomised controlled trials such as the Women’s Health Initiative are the best type of study to look at cause and effect, but typically are smaller and have a limited follow-up period. It is unusual for these types of studies to be used specifically to look at long-term safety outcomes, especially if the absolute risk of the outcome is low.
  • Prospective cohort studies such as the Million Women Study can follow people up for much longer. However, in this type of study people have not been randomly assigned to receive treatment or a control. This means that people who have received HRT could have different characteristics or past exposures compared to people who have not received HRT, and this may affect the likelihood of developing breast cancer.

What did the research involve?

The nine criteria the researchers looked at were:

  • Time order: did the women develop breast cancer after they had received HRT?
  • Information bias: does anxiety about the possibility that HRT may cause breast cancer influence how the women responded to questions in the study?
  • Detection bias: when HRT is prescribed, women are advised to have regular breast examinations and mammograms – did differences in the frequency of testing increase the number of women whose breast cancer would be detected?
  • Confounding: were the women who took HRT in the studies more likely to be exposed to other factors which may affect their risk of developing breast cancer?
  • Statistical stability and strength of association: how robust were the statistics? Was the risk of breast cancer very different or of borderline significance?
  • Dose-duration response: does being on a higher dose for a longer period of time increase the risk further?
  • Internal consistency: how variable were the risks in the study population? For example, were there any differences between women in terms of baseline characteristics such as age or weight?
  • External consistency: did the study’s findings agree with other similar studies?
  • Biological plausibility: is there a potential mechanism by which the hormones in HRT could trigger breast cancer in post-menopausal women?

What were the basic results?

The researchers suggested that the Collaborative Reanalysis study did not adequately satisfy the criteria of time order, bias, confounding, statistical stability and strength of association, dose-duration response, internal consistency, external consistency and biological plausibility.

The Women’s Health Initiative looked at two types of HRT: oestrogen-only HRT and combined oestrogen and progesterone HRT. Analysis of the trial of oestrogen plus progesterone HRT found that the findings did not adequately satisfy the criteria or bias, confounding, statistical stability and strength of association, duration response, internal consistency, external consistency or biological plausibility to suggest a causal association. The oestrogen-only HRT trial found that there was no increased risk of breast cancer in women taking this type of HRT relative to the control group.

The researchers said that despite the large size of the Million Women Study (which had followed women receiving oestrogen-only HRT, oestrogen and progesterone HRT, or no HRT) the study failed to satisfy several of the causality criteria. These included time order, information bias, detection bias confounding, statistical stability and strength of association, duration-response, internal consistency, external consistency or biological plausibility.

How did the researchers interpret the results?

The researchers concluded that HRT may or may not increase the risk of breast cancer and none of the three studies is able to establish that it does. The researchers also said that the randomised controlled trial of oestrogen-only HRT had shown no increased risk of breast cancer.


Researchers have reanalysed the design and data from three studies that had suggested that HRT is associated with an increased risk that breast cancer would develop. They wanted to determine whether HRT caused breast cancer to develop (that the link was “causal”). The researchers looked at a pooled analysis study called the Collaborative Reanalysis, the Women’s Health Initiative randomised controlled trials and the Million Women Study, which was a large prospective cohort study including 800,000 post-menopausal women.

The researchers found that these studies each failed to meet the majority of nine criteria which would be need to be met in order to say whether the studies could establish causality. These criteria included whether the women in the studies took HRT before their cancer developed. They looked at whether the studies had controlled for any confounding factors (where a factor may be associated with both the likelihood of taking HRT and the likelihood of developing breast cancer). The researchers also looked at the biological plausibility of any causal link.

This research is important because it makes us reconsider the evidence that suggested HRT might increase risk of breast cancer rather than there just being an association between the two. This new research shows that these studies can only tell us that there may or may not be a link between HRT and increased risk of breast cancer. Further, carefully designed studies are needed to determine whether HRT does indeed cause an increased risk of breast cancer.

Women who are taking HRT for menopausal symptoms or who used to take HRT should note the advice from Cancer Help UK, which states that any cancer risk associated with HRT reduces to a normal level five years after you stop taking HRT.

  • Comment

Have your say

You must sign in to make a comment

Please remember that the submission of any material is governed by our Terms and Conditions and by submitting material you confirm your agreement to these Terms and Conditions. Links may be included in your comments but HTML is not permitted.